Encrypted login | home

Program Information

Clinical Comparison of An IMRT Plan Computed with Four Algorithms

no image available
J Barbiere

J Barbiere*, A Ndlovu , Hackensack University Medical Center, Hackensack, NJ


SU-F-T-400 (Sunday, July 31, 2016) 3:00 PM - 6:00 PM Room: Exhibit Hall

Purpose:This work demonstrates clinical differences in a prostate IMRT plan DVH depending on which of four common algorithms is used for computation. Previous comparisons were often based on simple static or re-optimized IMRT plans which may not be clearly indicative of the clinical end result due solely to the computation algorithm.

Methods:A step and shoot prostate multiple field IMRT plan was created in Pinnacle with a Collapsed Cone Convolution Superposition (CCCS) algorithm. The computed patient plan with corresponding dose and the control points without dose were both independently exported to Eclipse. Using the same patient image and structure data sets the control points were recomputed with the Anisotropic Analytical Algorithm (AAA) and Acuros XB for dose-to-water in medium (AXBwater) and dose-to-medium in medium (AXBmedium). All DVH values were measured in Eclipse to eliminated known differences between various DVH systems. A 0.3cm3 volume was created at the isocenter for point dose (PD) comparison.

Results:The prostate target Acuros dose-to-medium in medium is clearly different than the others with a mean value of 101.7 cGy versus 103.8, 103.5 and 103.1 cGy for AAA, CCCS and AXBwater respectively. The difference in other clinical parameters such as volume at 105% dose (V105) is also unexpectedly different with values of 1.6, 23.9, 16.0 and 11.6 % respectively. The mean value in a small volume around the isocenter was 99.6, 102.4, 102.2 and 101.5 respectively which though not a great difference could affect the plan normalization and add to other effects in all regions of interest. The V102 is 25% larger for AAA compared to AXBwater. The variation in V50 for the bladder and rectum was approximately 5%.

Conclusion:This work demonstrates how an individual plan can be recomputed and compared directly between various algorithms and that clinical parameters can indeed vary according to which algorithm is used for computation.

Contact Email: